public XElement ParagraphsToXml(Paragraphs paragraphs, TableBuilder tableBuilder, Object builder = null)
{
var lop = paragraphs.Cast <Paragraph>().ToList();
return(ParagraphsToXml(lop, tableBuilder, builder));
}
/// <summary>
/// Takes the list of paragraphs and returns the XML representation of these elements.
/// </summary>
/// <param name="paragraphs">The list of paragraphs to parse.</param>
/// <param name="tableBuilder"></param>
/// <param name="builder"></param>
/// <returns>An XElement representation of the document.</returns>
public XElement ParagraphsToXml(List <Paragraph> paragraphs, TableBuilder tableBuilder, Object builder = null)
{
SidebarArticleParser = new SidebarArticleParser();
WdListType previouslistType = paragraphs[0].Range.ListFormat.ListType;
bool previousWasList = false;
string previousIFrameStyle = string.Empty;
bool previousWasIFrame = false;
bool desktopCodeFound = false;
bool currentIframeIsInsecure = false;
bool currentContainsInvalidNodes = false;
List <string> insecureIFrames = new List <string>();
bool containsInvalidNodes = false;
string iframeGroupId = String.Empty;
XElement iframeGroupElement = new XElement("div");
iframeGroupElement.SetAttributeValue("class", "iframe-component");
bool previousWasBlockquote = false;
var contiguousListElements = new List <Paragraph>();
var contiguousBlockquoteElements = new List <Paragraph>();
List <string> StylesToIgnore = ArticleDocumentMetadataParser.GetInstance().MetadataStyles;
Paragraph lastParagraph = null;
//var xData = new XElement("root");
var xData = new XElement("div");
xData.SetAttributeValue("class", "root");
Errors = new List <string>();
int imageTagCount = 0;
XElement divElement = null;
for (int i = 0; i < paragraphs.Count; i++)
{
Paragraph paragraph = paragraphs[i];
int tIndex = tableBuilder == null ? -1 : tableBuilder.GetTableIndexFor(paragraph.Range);
Style style = (Style)paragraph.get_Style();
if (tableBuilder != null && tIndex != -1 && tableBuilder.HasRetrieved(tIndex))
{
continue;
}
var currentStyle = (Style)paragraph.get_Style();
if (StylesToIgnore.Contains(currentStyle.NameLocal))
{
continue;
}
XElement xElement = null;
bool isList = paragraph.Range.ListParagraphs.Count > 0;
bool isBlockquote = currentStyle.NameLocal == BlockquoteName;
WdListType currentListType = paragraph.Range.ListFormat.ListType;
if (tableBuilder != null && tIndex != -1 && !tableBuilder.HasRetrieved(tIndex))
{
xData.Add(tableBuilder.ParseTable(tIndex));
continue;
}
if (builder != _quickFactsSideboxParser && _quickFactsSideboxParser != null &&
_quickFactsSideboxParser.Match(currentStyle.NameLocal))
{
_quickFactsSideboxParser.Add(paragraph);
continue;
}
if (_quickFactsSideboxParser != null && !_quickFactsSideboxParser.Match(currentStyle.NameLocal) &&
!_quickFactsSideboxParser.IsEmpty())
{
xData.Add(_quickFactsSideboxParser.GetSidebox(this));
_quickFactsSideboxParser.Clear();
}
lastParagraph = paragraph;
if (currentStyle.NameLocal == SidebarArticleParser.SidebarStyle)
{
try
{
xData.Add(paragraph.Range.Text);
SidebarArticleParser.RetrieveSidebarToken(paragraph);
}
catch (ArgumentException e)
{
Errors.Add(e.Message);
}
continue;
}
if (!isList && previousWasList)
{
xData.Add(GetListStyleElement(contiguousListElements));
contiguousListElements = new List <Paragraph>();
previousWasList = false;
}
if (!isBlockquote && previousWasBlockquote)
{
xData.Add(BlockquoteTransformer.Generate(contiguousBlockquoteElements, CharacterStyleTransformer));
contiguousBlockquoteElements = new List <Paragraph>();
previousWasBlockquote = false;
}
if (isBlockquote)
{
contiguousBlockquoteElements.Add(paragraph);
previousWasBlockquote = true;
//just forming contiguous blocks, no further processing
continue;
}
if (isList)
{
if (previouslistType != currentListType && contiguousListElements.Count > 0)
{
xElement = GetListStyleElement(contiguousListElements);
xData.Add(xElement);
contiguousListElements = new List <Paragraph>();
}
//add paragraph to bloc,
//even if new list has been started, we are adding it now
contiguousListElements.Add(paragraph);
previousWasList = true;
previouslistType = paragraph.Range.ListFormat.ListType;
//just forming contiguous blocks, no further processing
continue;
}
if (style.NameLocal == DocumentAndParagraphStyles.IFrameCodeStyle ||
style.NameLocal == DocumentAndParagraphStyles.IFrameMobileCodeStyle)
{
var iframeElement = new XElement("div");
if (!previousWasIFrame)
{
iframeGroupId = Guid.NewGuid().ToString("N");
previousWasIFrame = true;
iframeGroupElement.SetAttributeValue("class", "iframe-component");
}
string cssStyle = string.Empty;
if (style.NameLocal == DocumentAndParagraphStyles.IFrameCodeStyle)
{
desktopCodeFound = true;
cssStyle = String.Format("ewf-desktop-iframe_{0}", iframeGroupId);
iframeElement.SetAttributeValue("class", $"iframe-component__desktop {cssStyle}");
iframeElement.SetAttributeValue("data-mediaid", iframeGroupId);
}
if (style.NameLocal == DocumentAndParagraphStyles.IFrameMobileCodeStyle)
{
cssStyle = String.Format("ewf-mobile-iframe_{0}", iframeGroupId);
iframeElement.SetAttributeValue("class", $"iframe-component__mobile {cssStyle}");
}
var insecureIFramesInParagraph = HTMLTools.CheckForInsecureIFrames(paragraph.Range.Text);
currentIframeIsInsecure = insecureIFramesInParagraph.Any() || currentIframeIsInsecure;
insecureIFrames.AddRange(insecureIFramesInParagraph);
currentContainsInvalidNodes = HTMLTools.ContainsForExternalNodes(paragraph.Range.Text);
containsInvalidNodes = containsInvalidNodes || currentContainsInvalidNodes;
XElement embedElement = IFrameEmbedBuilder.Parse(paragraph, cssStyle, true);
if (embedElement != null)
{
iframeElement.SetAttributeValue("data-embed-link", "enabled");
iframeElement.Add(embedElement);
iframeGroupElement.Add(iframeElement);
}
if (style.NameLocal == DocumentAndParagraphStyles.IFrameMobileCodeStyle)
{
xData.Add(iframeGroupElement);
iframeGroupElement = new XElement("div");
iframeGroupElement.SetAttributeValue("class", "iframe-component");
}
previousIFrameStyle = style.NameLocal;
continue;
}
if (IFrameEmbedBuilder.IFrameStyles.Contains(style.NameLocal))
{
WordStyleStruct w = new WordStyleStruct();
//base styles are used becuase the parent level styles only exist in the plugin
var baseStyle = (Style)style.get_BaseStyle();
if (baseStyle != null)
{
ParagraphStyles.TryGetValue(baseStyle.NameLocal, out w);
}
XElement curElement = new XElement("p");
string stylesection = String.Empty;
bool isNewIframe = false;
if (style.NameLocal == DocumentAndParagraphStyles.IFrameHeaderStyle)
{
stylesection = "header";
isNewIframe = true;
}
if (style.NameLocal == DocumentAndParagraphStyles.IFrameTitleStyle)
{
stylesection = "title";
isNewIframe = previousIFrameStyle != DocumentAndParagraphStyles.IFrameHeaderStyle;
}
if (style.NameLocal == DocumentAndParagraphStyles.IFrameCaptionStyle)
{
stylesection = "caption";
isNewIframe = previousIFrameStyle != DocumentAndParagraphStyles.IFrameMobileCodeStyle;
}
if (style.NameLocal == DocumentAndParagraphStyles.IFrameSourceStyle)
{
stylesection = "source";
isNewIframe = previousIFrameStyle != DocumentAndParagraphStyles.IFrameCaptionStyle &&
previousIFrameStyle != DocumentAndParagraphStyles.IFrameMobileCodeStyle;
}
if (!previousWasIFrame && isNewIframe)
{
iframeGroupId = Guid.NewGuid().ToString("N");
previousWasIFrame = true;
}
string cssClass = w == null ? string.Empty : w.CssClass;
string iframeIdClass = String.Format("{0}-{1}_{2}", IFrameEmbedBuilder.IFrameClassName, stylesection,
iframeGroupId);
curElement.SetAttributeValue("class",
String.Format("{0} {1}",
cssClass,
iframeIdClass));
curElement = CharacterStyleTransformer.GetCharacterStyledElement(curElement, paragraph,
CharacterStyleFactory.GetCharacterStyles(), false);
xData.Add(curElement);
previousIFrameStyle = style.NameLocal;
continue;
}
else //redundant else but adds code clarity
{
if (currentIframeIsInsecure || currentContainsInvalidNodes ||
(!desktopCodeFound && previousWasIFrame))
{
xData.Elements()
.Where(x => x.Attribute("class") != null &&
x.Attribute("class").Value.Contains(iframeGroupId))
.Remove();
}
previousIFrameStyle = String.Empty;
previousWasIFrame = false;
currentIframeIsInsecure = false;
currentContainsInvalidNodes = false;
desktopCodeFound = false;
}
if (ImageReferenceBuilder.Parse(paragraph) != null)
{
if (imageTagCount == 0)
{
divElement = new XElement("section");
divElement.SetAttributeValue("class", "article-exhibit");
xData.Add(divElement);
}
//Get the float Value from the image hyperlink (if it is an image) and set it to the article-image element
var hyprlnk = paragraph.Range.Hyperlinks.Cast <Hyperlink>().FirstOrDefault();
if (hyprlnk != null && string.IsNullOrEmpty(hyprlnk.ScreenTip) == false)
{
string classValue;
classValue = imageFloatDictionary.TryGetValue(hyprlnk.ScreenTip.ToLower(), out classValue) ? classValue : string.Empty;
divElement.SetAttributeValue("class", classValue);
}
var imageTag = ImageReferenceBuilder.Parse(paragraph);
divElement.Add(imageTag);
imageTagCount++;
continue;
}
imageTagCount = 0;
WordStyleStruct styleStruct;
if (ParagraphStyles.TryGetValue(currentStyle.NameLocal, out styleStruct))
{
//if there is a special configuration for the paragraph style, have it configured properly
string element = styleStruct.CssElement;
if (element.IsNullOrEmpty())
{
element = null;
}
string clas = styleStruct.CssClass;
if (clas.IsNullOrEmpty())
{
clas = null;
}
XElement curElement = element != null ? new XElement(element) : new XElement("p");
if (!string.IsNullOrWhiteSpace(clas))
{
curElement.SetAttributeValue("class", clas);
}
xElement = CharacterStyleTransformer.GetCharacterStyledElement(curElement, paragraph,
CharacterStyleFactory.GetCharacterStyles(), false);
//The next section is group multiple paragraphs belonging to the same answer under the same Answer element to prevent multiple answer styling on front-end for a single answer
if (clas == "article-interview__answer")
{
while (paragraphs.Count > i + 1)//If there are still more element to inspect
{
//Get the styling of the next element
var tempAnswerStyle = (Style)paragraphs[i + 1].get_Style();
WordStyleStruct tempAnswerStyleStruct;
//Get the cssClass of the next element
ParagraphStyles.TryGetValue(tempAnswerStyle.NameLocal, out tempAnswerStyleStruct);
//if it is also an answer paragraph
if (tempAnswerStyleStruct.CssClass == clas)
{
//add the the next paragraph content into the current answer body
xElement.Add(CharacterStyleTransformer.GetCharacterStyledElement(new XElement("p"), paragraphs[i + 1], CharacterStyleFactory.GetCharacterStyles(), false));
i++;
}
else
{
break;
}
}
}
}
else
{
xElement = new XElement("p");
xElement = CharacterStyleTransformer.GetCharacterStyledElement(xElement, paragraph,
CharacterStyleFactory.GetCharacterStyles(), false);
}
xData.Add(xElement);
previouslistType = paragraph.Range.ListFormat.ListType;
}
if (insecureIFrames.Any())
{
var confirmSave =
MessageBox.Show("You have inserted multimedia items using non-secure links. " +
"Click 'OK' to continue saving the article without this content or 'Cancel'" +
" to go back into the article and edit your multimedia items.",
"Continue with save?",
MessageBoxButtons.OKCancel, MessageBoxIcon.Exclamation);
if (confirmSave != DialogResult.OK)
{
throw new InsecureIFrameException(insecureIFrames);
}
}
if (containsInvalidNodes)
{
var confirmSave =
MessageBox.Show("You have inserted multimedia items with invalid or non-permitted HTML code. " +
"Click 'OK' to continue saving the article without this content or 'Cancel' to " +
"go back into the article and edit your multimedia items.", "Continue with save?",
MessageBoxButtons.OKCancel, MessageBoxIcon.Exclamation);
if (confirmSave != DialogResult.OK)
{
throw new InvalidHtmlException();
}
}
if (lastParagraph != null && lastParagraph.Range.ListParagraphs.Count > 0)
{
xData.Add(GetListStyleElement(contiguousListElements));
}
if (lastParagraph != null && ((Style)lastParagraph.get_Style()).NameLocal == BlockquoteName)
{
xData.Add(BlockquoteTransformer.Generate(contiguousBlockquoteElements, CharacterStyleTransformer));
}
if (_quickFactsSideboxParser != null && builder != _quickFactsSideboxParser &&
!_quickFactsSideboxParser.IsEmpty())
{
xData.Add(_quickFactsSideboxParser.GetSidebox(this));
_quickFactsSideboxParser.Clear();
}
return(xData);
}
//private float TableWidth(Table table)
//{
// int colIndex = 1;
// float width = 0;
// try
// {
// while (true)
// { //major hacks because documentation is lacking
// and i can't find another solution
// table.Rows is inaccessble if there are merged cells
// across rows
// table.Columns is inaccessible if there are merged
// cells across columns
// so how do i find out the number of cells in a table row?
// WE KEEP GOING TILL WE BUST
// width += table.Cell(1, colIndex).Width;
// colIndex++;
// }
// } catch
// {
// return width;
// }
//}
public XElement ParseTable(int index)
{
Table table = GetTable(index);
_retrieved[index] = true;
var root = new XElement("table");
root.SetAttributeValue("class", "data");
var tbody = new XElement("tbody");
root.Add(tbody);
//var rows = table.Rows;
var wordUtils = new WordUtils();
var tableAnalyzer = new TableAnalyzer(table);
const int maxTableWidth = 544;
decimal curSetWidth = Math.Truncate(tableAnalyzer.TableCellWidthSums.Last());
if (curSetWidth < maxTableWidth)
{
root.SetAttributeValue("width", curSetWidth);
}
bool first = true;
for (int r = 1; r <= table.Rows.Count; r++)
{
//Row row = rows[r];
var currentRow = new XElement("tr");
tbody.Add(currentRow);
//var rowCellEnumerator = row.Cells.Cast<Cell>().GetEnumerator();
decimal currentWidthSum = 0;
int numColSpansUsed = 0;
//while (rowCellEnumerator.MoveNext())
for (int c = 1; c <= table.Columns.Count; c++)
{
try
{
var curCell = table.Cell(r, c); //rowCellEnumerator.Current);
curCell.Select();
var app = curCell.Application;
string cellType = first ? "th" : "td";
var xmlCell = new XElement(cellType);
currentRow.Add(xmlCell);
Paragraphs paragraphs = curCell.Range.Paragraphs;
var tableBuilder = new TableBuilder(curCell.Tables);
XNode currentDescendent = wordUtils.ParagraphsToXml(paragraphs, tableBuilder).FirstNode;
while (currentDescendent != null)
{
xmlCell.Add(currentDescendent);
currentDescendent = currentDescendent.NextNode;
}
currentWidthSum += (decimal)curCell.Width;
int widthIndex = tableAnalyzer.TableCellWidthSums.IndexOf(currentWidthSum);
int curNumCol = widthIndex + 1 - numColSpansUsed;
xmlCell.SetAttributeValue("colspan", curNumCol);
xmlCell.SetAttributeValue("rowspan",
app.Selection.Information[WdInformation.wdEndOfRangeRowNumber] -
app.Selection.Information[WdInformation.wdStartOfRangeRowNumber] + 1);
decimal cellWidth = 0;
for (int i = widthIndex; i >= widthIndex - curNumCol + 1; i--)
{
cellWidth += tableAnalyzer.CellWidthPercentages[i];
}
xmlCell.SetAttributeValue("width", cellWidth + "%");
numColSpansUsed += curNumCol;
}
catch (System.Runtime.InteropServices.COMException)
{ //it seems that the only way to tell if a cell at a particular [row, col] index
//exists is to call table.Cell(row, col) and see if it throws an exception
//the reason it may not exist is rowspans and colspans
int nr = r;
while (nr > 0)
{
try
{
currentWidthSum += (decimal)table.Cell(nr, c).Width;
int widthIndex = tableAnalyzer.TableCellWidthSums.IndexOf(currentWidthSum);
int curNumCol = widthIndex + 1 - numColSpansUsed;
numColSpansUsed += curNumCol;
break;
}
catch (System.Runtime.InteropServices.COMException)
{
nr--;
}
}
}
//var colspan = (int)(curCell.Width / smallestWidth);
//if (colspan > 1) xmlCell.SetAttributeValue("colspan", colspan);
}
first = false;
}
return(root);
}
